Tag
This paper introduces Neural Optimal Particle Filters (NOPFs), which learn amortized proposal distributions to improve sample efficiency and distributional accuracy in sequential inference tasks, enhancing classical particle filters with data-driven methods.